JP2002247413A - Image noise removal device - Google Patents

Abstract

(57) [Summary] [Problem] To mainly remove noise on the background without losing contours and subtle shadows of the subject.
Provided is a video noise elimination device in which noise does not appear to increase or decrease periodically. A variable limiter detects noise in a luminance signal according to a threshold value determined by a horizontal luminance edge / average extraction circuit, a vertical correlation extraction circuit, and a skin color detection circuit according to characteristics of an input video signal. The detected noise is attenuated by the variable attenuator 10 according to the attenuation amount similarly determined, and the first subtracter 7 subtracts the noise from the luminance signal to remove noise included in the luminance signal.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for reproducing a video signal by recording video information on a tape medium or a disk medium or transmitting video information as a color difference signal such as a satellite broadcast or a terrestrial broadcast. The present invention relates to a video noise removing device for removing video noise.

[0002]

2. Description of the Related Art Conventionally, video output recorded on a tape medium or a disk medium or transmitted from a satellite broadcast, a cable broadcast or a terrestrial broadcast is usually transmitted as a composite video signal so that it can be reproduced by a television receiver. In recent years, MPEG
2. Description of the Related Art With the spread of video equipment to which a compression technique is applied, a video noise elimination device that transmits a video signal using a color difference signal and mainly removes video noise in a background portion of a video is being introduced.

FIG. 9 is a block diagram showing, as an example of the prior art relating to the present invention, a configuration of a video noise eliminator for reproducing a color difference video signal recorded on a disc. In FIG. 9, reference numeral 1 denotes a disk on which a color difference video signal is encoded in a signal form suitable for recording (or reproduction), modulated, and recorded. A pickup 2 converts an information signal recorded on the disk 1 into an electric signal. Reference numeral 3 denotes a disk rotating device which rotates the disk 1 at a rotation speed suitable for reproduction. 4 is a video signal reproducing circuit,
The video signal recorded on the disk 1 is demodulated and decoded, and output as a luminance signal and a color difference video signal. Reference numeral 5 denotes a first band pass filter, which passes a band around 3 MHz of the luminance signal in the output of the video signal reproducing circuit 4. Reference numeral 6 denotes a first limiter circuit which compares the output of the first band-pass filter 5 with a threshold value and outputs the result after limiting. 7 is the first
From the luminance signal output of the video signal reproducing circuit 4
The output of the first limiter 6 is subtracted and output. Reference numeral 8 denotes an encoder which monitors an interlaced scanning video signal (not shown).
The video signal is converted to a video signal suitable for output. Reference numeral 9 denotes a video signal output terminal, and a video output reproduced from the video signal output terminal is output to a monitor (not shown).

[0004] The operation of the conventional video noise eliminator configured as described above will be described.

When the disk 1 is being rotated at a rotational speed suitable for reproduction by the disk rotating device 3, the pickup 2 reads an optical signal recorded on the disk 1, converts the optical signal into an electric signal, and converts the optical signal into a video signal. The video signal reproducing circuit 4 provided to the reproducing circuit 4 converts the supplied electric signal into a digital video signal, and outputs the digital video signal to the first subtractor 7 and the first band pass filter 5.

FIG. 10 is a diagram showing characteristics of a first band pass filter of a conventional video noise elimination device.

As shown in FIG. 10, the first band-pass filter 5 has a luminance signal frequency band of 6 MHz,
Among them, it has a characteristic of passing only around 3 MHz where the image noise component is large.

[0008] The luminance signal that has passed through the first band-pass filter 5 is then input to a first limiter 6. FIG.
FIG. 4 is a diagram showing characteristics of a first limiter of a conventional video noise elimination device. As shown in FIG. 11, the absolute value of the input is 0
Until a certain value X ₀ from the first limiter causes all input passage, the value X _0, until the value 2X ₀ value 2X ₀ - outputs the input, if the absolute value than the value 2X ₀ Outputs 0.

This functions to extract only the minute amplitude components contained in the video signal as video noise. Now the first
Among the luminance signals extracted only in the 3 MHz band by the band pass filter 5, only the minute amplitude component is extracted as luminance signal noise by the above-described method.

The luminance signal noise thus extracted is subtracted from the luminance signal by the first subtractor 7 to remove the luminance signal noise, and is sent to the encoder 8.

The encoder 8 converts the digital video signal reproduced and noise-removed as described above into a video signal suitable for outputting to a monitor (not shown). That is,
A synchronizing signal is added to the digitally input luminance signal, and the signal is output as an analog signal. The two color difference signals CB and CR are also converted into analog signals and output. The video signal recorded on the disc 1 is progressive (48
In the case of the (0P) system, it is also output as a progressive color difference signal. Further, a chrominance signal modulated by a chrominance subcarrier is created from the two signals CB and CR, and a chrominance signal, a luminance signal, and a synchronizing signal are combined to generate a composite video signal. The analog luminance signal, color difference signal, color signal, and composite video signal generated in this manner are output from the video signal output terminal 9.

FIG. 12 is a diagram showing noise removal characteristics of a conventional video noise removal device. The luminance input (a) in FIG. 12 is a luminance signal output of the video signal reproducing circuit 4 in FIG.
Of this waveform, a portion having relatively high luminance is a main subject portion, and a portion having relatively low luminance is a background portion. When such a signal passes through the first band-pass filter 5, a waveform as shown in FIG. 12 (b) first band-pass filter output is obtained. Further, when the signal passes through the first limiter 6, a waveform shown in FIG. 12 (c) noise output is obtained. Although only the small amplitude component is extracted here, the boundary region of the part where the transition from the background to the subject is not noise, but the amplitude is extracted as noise before and after the limiter is applied. I have.

As described above, when a noise signal in which a signal which is not originally noise is extracted as noise is subtracted from the luminance input of FIG. 12A, a waveform as shown in FIG. 12D luminance output is obtained. Become.

That is, in the conventional video noise elimination apparatus, when mainly removing background noise, the background noise is eliminated, and at the same time, the contour is blurred because the signal of the outline portion is blunted, and the subject is also removed. The minute amplitude of the video signal is lost and subtle shadows cannot be reproduced.

FIG. 13 is a diagram showing a time change characteristic of noise of the conventional video noise removing apparatus. In the conventional video noise elimination device, a video signal encoded by the MPEG encoding method is recorded on the disk 1. In the (q) coding method shown in FIG. 13, each video frame is coded by any of the coding methods of intra-frame coding (I), forward prediction coding (P), and bidirectional prediction coding (B). Indicates whether it is information. Here, in intra-frame encoding (I), compression encoding is performed using only intra-frame information of the original image, whereas in forward prediction encoding (P), motion prediction is performed from a past frame. In the bidirectional prediction encoding (B), encoding is performed with difference information from past and future frame information. Therefore, as shown in FIG. 13 (r), the information amount of the compressed frame is the largest in the intra-frame coded (I) frame and the smallest in the bidirectional predictive coded (B) frame. Conversely, as shown in the visual noise amount of the reproduced image in FIG. 13 (s), in the intra-coded (I) frame, while there is much information on components visually detected as noise, In bidirectional predictive coding (B) frames,
There is little information on components visually detected as noise.

Therefore, as shown in FIG.
As shown in the visual noise amount after the noise removal, the residual noise after the noise removal changes with time. In the MPEG encoding method, since an intra-coded (I) frame appears periodically, a phenomenon occurs in which noise appears to increase or decrease periodically. This is particularly noticeable when the amount of noise removal is smaller than the amount of noise.

[0017]

As described above, the conventional video noise elimination apparatus has a problem in that, when mainly removing noise on the background, deterioration of the contour portion and elimination of the minute amplitude of the subject are eliminated. have. There is also a problem that the noise appears to increase or decrease periodically. In the future, it will be possible to remove noise in the background easily and inexpensively, without removing the deterioration of the contour and the minute amplitude of the subject, and without the noise appearing to increase or decrease periodically. Equipment installation is required.

The present invention solves the above-mentioned problems of the prior art, and includes a noise extracting means for detecting noise of a luminance signal according to a threshold value determined according to characteristics of an input video signal,
Attenuating means for attenuating the output of the noise extracting means according to the amount of attenuation determined according to the characteristics of the input video signal, and noise removing means for subtracting the output of the attenuating means from the luminance signal and removing noise contained in the luminance signal The noise extraction sensitivity and the amount of noise removal are changed according to various information of the video signal, without eliminating the deterioration of the contour and the minute amplitude of the subject.
It is another object of the present invention to provide an image noise elimination apparatus capable of mainly removing background noise without the noise appearing to increase or decrease periodically.

[0019]

SUMMARY OF THE INVENTION In order to solve this problem, the present invention provides a noise extracting means for detecting noise of a luminance signal according to a threshold value determined according to the characteristics of an input video signal; An image attenuating means for attenuating the output of the noise extracting means in accordance with the amount of attenuation determined according to the characteristics of the above, and a noise removing means for subtracting the output of the attenuating means from the luminance signal to remove noise contained in the luminance signal It is a noise removing device.

Thus, according to various information of the video signal,
Since the noise extraction sensitivity and the noise removal amount are changed, it is possible to mainly remove the noise in the background without deteriorating the contour and eliminating the minute amplitude of the subject.

In particular, the threshold value of the noise extracting means and the attenuation of the attenuating means depend on whether the video signal frame is an intra-frame encoding, a forward predictive encoding, or a bidirectional predictive encoding. By switching both or any one of the above, the noise mainly in the background can be removed without the noise appearing to increase or decrease periodically.

Further, the noise extraction means includes a band-pass filter means for passing only a specific frequency component of the luminance signal, and a limiter means for comparing the output of the band-pass filter means with a threshold value and passing the signal only below the threshold value. The noise extraction sensitivity and the
By changing the noise removal amount, it is possible to remove mainly the noise in the background without eliminating the deterioration of the contour and the minute amplitude of the subject, and without the noise appearing to increase or decrease periodically. .

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (Embodiment 1) Hereinafter, Embodiment 1 of the present invention will be described with reference to FIGS. 1 to 8 in the case of reproducing a signal recorded on a disc.

FIG. 1 is a block diagram showing a configuration of a video noise removing apparatus according to Embodiment 1 of the present invention. In FIG. 1, reference numeral 1 denotes a disk on which a color difference video signal is encoded in a signal form suitable for recording (or reproduction), modulated, and recorded. A pickup 2 converts an information signal recorded on the disk 1 into an electric signal. Reference numeral 3 denotes a disk rotating device which rotates the disk 1 at a rotation speed suitable for reproduction. Reference numeral 4 denotes a video signal reproducing circuit which demodulates and decodes the video signal recorded on the disk 1 and outputs it as a luminance signal and a color difference video signal. Reference numeral 5 denotes a first band pass filter, which allows a band around 3 MHz of the luminance signal of the output of the video signal reproducing circuit 4 to pass. Reference numeral 7 denotes a first subtracter which subtracts the output of the first limiter 6 from the luminance output of the video signal reproducing circuit 4 and outputs the result. Reference numeral 8 denotes an encoder that converts an interlaced scanning video signal into a video signal suitable for a monitor (not shown) and outputs the video signal. Reference numeral 9 denotes a video signal output terminal, and a video output reproduced from the video signal output terminal is output to a monitor (not shown). 1
Numeral 0 denotes a variable attenuator, which can change the amount of attenuation by an external control input, and attenuates the variable limiter output 16 to provide the first subtractor 7
Output to Reference numeral 11 denotes a horizontal luminance feature extraction circuit, which outputs a horizontal average luminance level and contour information of the luminance signal. 1
Reference numeral 2 denotes a threshold setting circuit that controls the threshold of the variable limiter 16. Reference numeral 13 denotes an encoding type extraction circuit, which indicates which of the reproduced video signal frame is one of intra-frame encoding, forward prediction encoding, and bidirectional prediction encoding. Reference numeral 14 denotes a vertical correlation extracting circuit which detects and outputs the vertical correlation of the luminance signal. A skin color detection circuit 15 detects and outputs a skin color area from the color difference signal. Reference numeral 16 denotes a variable limiter circuit which compares the output of the first bandpass filter 5 with a threshold value and outputs the result after limiting.

The operation of the video noise elimination apparatus according to the first embodiment of the present invention configured as described above will be described.

When the disk 1 is being rotated by the disk rotating device 3 at a rotational speed suitable for reproduction, the pickup 2 reads the optical signal recorded on the disk 1 and converts it into an electrical signal to convert it into a video signal. It is given to the reproduction circuit 4. The video signal reproducing circuit 4 converts the applied electric signal into a digital color difference signal and outputs the digital color difference signal to the first subtractor 7 and the first band pass filter 5.

FIG. 2 is a diagram showing characteristics of the first band-pass filter of the video noise elimination device according to the first embodiment of the present invention.

As shown in FIG. 2, the first band-pass filter 5 has a frequency band of a luminance signal of 6 MHz, and has a characteristic of passing only around 3 MHz having a large image noise component. The luminance signal that has passed through the first band-pass filter is then input to the variable limiter 16.

FIG. 3 is a diagram showing characteristics of the variable limiter of the video noise elimination device according to the first embodiment of the present invention.

[0030] As shown in FIG. 3, until the absolute value of the input is the value X _n (n = 1, 2, 3) with 0, the first limiter is passed through all of the input from the values X _n , until the value 2X _n value 2X _n - outputs the input, if the absolute value is greater than the value 2X _n outputs 0. This serves to extract only the minute amplitude component contained in the video signal as video noise.

Now, the first band pass filter 5 outputs 3 MHz.
Of the luminance signals extracted only in the z band, only the minute amplitude components are extracted as luminance signal noise by the above-described method. As described later, n is set so that the noise extraction gain can be switched from the outside. When n = 0, n does not operate as a limiter and nothing is output.

The luminance signal noise extracted in this manner is input to the variable attenuator 10. The variable attenuator 10 attenuates the input output of the variable limiter 16, that is, the detected noise signal by a multiplier determined by an external control signal and outputs the signal. The output of the variable attenuator 10 is the first
The noise of the luminance signal is removed by subtracting the luminance signal from the luminance signal by the subtracter 7 and sent to the encoder 8.

The encoder 8 converts the digital video signal reproduced and noise-removed as described above into a video signal suitable for outputting to a monitor (not shown). That is,
A synchronizing signal is added to the digitally input luminance signal, and the signal is output as an analog signal. The two color difference signals CB and CR are also converted into analog signals and output. The video signal recorded on the disc 1 is progressive (48
In the case of the (0P) system, it is also output as a progressive color difference signal. Further, a chrominance signal modulated by a chrominance subcarrier is created from the two signals CB and CR, and a chrominance signal, a luminance signal, and a synchronizing signal are combined to generate a composite video signal. The analog luminance signal, color difference signal, color signal, and composite video signal generated in this manner are output from the video signal output terminal 9.

FIG. 4 is a block diagram showing a configuration of the horizontal luminance edge / average extraction circuit 11 of the video noise elimination device according to the first embodiment of the present invention.

In FIG. 4, reference numeral 31 denotes a luminance input terminal from which a luminance signal is input. Reference numeral 32 denotes a first delay circuit, which is composed of four delay elements. Reference numeral 33 denotes a second delay circuit, which receives the output of the last-stage delay element of the first delay circuit 32 to provide a delay, and To the delay circuit 34 of FIG. Reference numeral 34 denotes a third delay circuit, which includes four delay elements. A first integrator 35 integrates the input of the luminance input terminal 31 and the output value of each delay element of the first delay circuit 32. 36 is a second integrator,
The output of the second delay circuit 33 and the output value of each delay element of the second delay circuit 34 are integrated. Reference numeral 37 denotes a second subtractor which outputs a difference between the first integrator 35 and the second integrator 36. 38 is a first threshold value. A first comparator 36 compares the value of the first threshold 38 with the value of the second subtractor 37 and outputs the result. Reference numeral 40 denotes an edge information output terminal from which an output of the first comparator 39 is output. An adder 41 adds the outputs of the first integrator 35 and the second integrator 36. 42 is a second threshold value. 43 is the second
And compares the value of the second threshold value 42 with the value of the adder 42 and outputs the result. Reference numeral 44 denotes an average luminance information output terminal from which an output of the second comparator 43 is output.

Now, the luminance signal output of the video signal reproducing circuit 4 of FIG. 1 is input to the luminance input terminal 31. In FIG. 4, the luminance signal input from the luminance input terminal 31 is the first luminance signal.
Is input to the delay circuit 32. The first delay circuit 32 has four delay elements, each of which delays and outputs one pixel time of luminance. The first integrator 35 calculates the sum of the luminance signal input and the outputs of the four delay elements. The output of the last-stage delay element of the first delay circuit 32 is
3 is input. That is, the output of the first integrator 35 corresponds to the average luminance level of five pixels following the luminance pixel data in the second delay circuit 33. Second delay circuit 33
Gives a delay of one pixel time to the luminance, and a third delay circuit 34
And the second integrator 36. Third delay circuit 34
Has four delay elements, each of which delays and outputs one pixel time of luminance. The second integrator 36 calculates the sum of the input of the third delay circuit 34 and the outputs of the four delay elements. That is, the output of the second integrator 36 corresponds to the average luminance level of the five pixels preceding the luminance pixel data in the second delay circuit 33.

The second subtractor 37 calculates the difference between the first integrator 35 and the second integrator 36. That is, when the pixel input to the second delay circuit 33 is the contour of the image, the average luminance values before and after are greatly different. In such a case, the absolute value of the output of the second subtractor 37 is used. Becomes larger, and if the pixel input to the second delay circuit 33 is not the outline of the image, the average luminance value before and after does not greatly differ. In such a case, the output of the second subtractor 37 is output. The absolute value becomes smaller. Therefore, by appropriately selecting the first threshold value 38, the first comparator 39 can detect whether the image is a contour portion or not. By setting the first threshold value 38 in this manner, output “1” is output from the edge information output terminal 40 in an image edge portion and “0” is output in a non-edge portion.

The adder 41 calculates the sum of the first integrator 35 and the second integrator 36. Therefore, the output of the adder 41 corresponds to the average luminance level of the peripheral pixels. As described above, the second threshold value 42 is set in order to obtain a portion where the average luminance is low and a portion where the average luminance is high, and the second comparator 42 compares the output of the adder 41 with the second threshold value 42 to obtain a pixel. Average luminance information corresponding to the peripheral average luminance is obtained and output from the average luminance information output terminal 44.

FIG. 5 is a block diagram showing a configuration of the vertical correlation extraction circuit 14 of the video noise elimination device according to the first embodiment of the present invention.

In FIG. 5, reference numeral 51 denotes a luminance input terminal from which a luminance signal is input. Reference numeral 52 denotes a line memory, which gives a delay corresponding to one horizontal scanning time to an input and outputs it. 53 is a fourth subtractor, which outputs the difference between the input and output of the line memory 52. Reference numeral 54 denotes a second band pass filter, which has a characteristic of passing a signal around 3 MHz. 55 is a third threshold value. Reference numeral 56 denotes a third comparator, which includes a value of the third threshold value 55 and the second band-pass filter 5.
4 is compared and output. Reference numeral 57 denotes a vertical correlation output terminal from which an output of the third comparator 56 is output.

Now, the luminance signal output of the video signal reproducing circuit 4 of FIG. 1 is input to the luminance input terminal 51. In FIG. 5, a luminance signal input from a luminance input terminal 51 is input to a line memory 52. The fourth subtractor 53 calculates the difference between the luminance information before and after the line memory 52. Since the delay time of the line memory is set to exactly one horizontal scanning time on the image, the output of the fourth subtractor 53 becomes a vertical correlation component. The output of the fourth subtractor 53 is input to the second bandpass filter 54, and only the 3 MHz peripheral band component is passed. That is, the second band-pass filter 54
Is a component of a peripheral band of 3 MHz which is uncorrelated in the vertical direction of the luminance information.

This component is a noise component that appears randomly in the vertical direction. Therefore, if the third threshold value 55 is set so that this component can be extracted, the third comparator 56 can classify the component around 3 MHz that is uncorrelated in the vertical direction, that is, a portion having noise and a portion not having noise. As described above, the signal of the portion having the vertical noise extracted by the third comparator 56 is output from the vertical correlation output terminal 57.

FIG. 6 is a diagram showing the operation of the skin color detection circuit 15 of the video noise elimination device according to the first embodiment of the present invention.

In FIG. 1, the color difference signal of the video signal reproducing circuit 4 is input to the skin color detecting circuit 15. The color difference signal is C
This is vector information of two axes B and CR. The angle of the vector indicates hue, and the absolute value indicates saturation. By detecting whether or not the vector of each pixel is included in the area indicated by the skin color area S in FIG. 6, it is possible to detect whether or not the image is a color close to the skin color. Figure 1
The skin color detection circuit 15 in this manner classifies and outputs the case where each pixel is close to the skin color (1) and the case where it is not (0).

In FIG. 1, a threshold setting circuit 12 controls the value of the threshold (n) of the variable limiter 16. Outputs of a horizontal luminance edge / average extraction circuit 11, a vertical correlation extraction circuit 14, a skin color detection circuit 15, and an encoding type extraction circuit 13 are connected to the threshold value setting circuit 12, and according to each output,
As shown in Table 1, a threshold is set. here,
The threshold value is a level at which nothing is output as 0, and as the numerical value increases as 1, 2, and 3, the noise detection level, that is, the variable limiter 16
Indicates that the limit value is high.

[0046]

[Table 1]

That is, noise is not detected at the edge portion (n = 0). In addition, since a portion having a high vertical correlation is highly likely to be a signal, the detection level is reduced (n = 1). In the non-edge and low vertical correlation portions, first, since the average luminance is high and the skin color portion is the subject, the noise detection level is reduced (n = 1) in order not to impair the resolution,
The detection level is set to medium (n = 2) because the part with high average luminance and non-skin color may be a subject of low interest or the background. Further, since the portion where the average luminance is low is the background, the noise component is most noticeable in the frame of the intra-frame encoding (I), so the detection level is increased (n
= 3), medium (n =
2), low (n = 1) in bidirectional predictive coding (B) frame
And

Next, the operation of the variable attenuator 10 will be described. The outputs of the horizontal luminance edge / average extraction circuit 11, the vertical correlation extraction circuit 14, the skin color detection circuit 15, and the encoding type extraction circuit 13 are connected to the variable attenuator 10. As shown, the gain is set. The variable attenuator 10 multiplies the output of the variable limiter 16 by a gain and outputs the result to the first subtractor 7.

[0049]

[Table 2]

That is, since the portion having a high vertical correlation is highly likely to be a signal, the gain is reduced (gain = 0.
3). In the portion having low vertical correlation, first, since the average luminance is high and the skin color portion is the subject, the gain is reduced (gain = 0.3) so as not to impair the resolution, and the average luminance is high and not the skin color. Since the part may be an object of low interest or the background, set the gain to medium (gain =
0.5). Further, since the portion where the average luminance is low is the background, the noise component is most conspicuous in the frame of the intra-frame encoding (I), so the gain is increased (gain =
1.0), the gain is set to medium (gain = 0.5) in the forward predictive coding (P) frame, and low (gain = 0.3) in the bidirectional predictive coding (B) frame.

FIG. 7 is a diagram showing noise elimination characteristics of the video noise elimination device according to the first embodiment of the present invention.

The (a) luminance input in FIG. 7 is the luminance output of the video signal reproducing circuit 4 in FIG. In this waveform, a portion having relatively high brightness is a main subject portion, and a portion having relatively low brightness is a background portion. 7 (e) shows an output of the line memory 52 in the vertical correlation extraction circuit 14. That is, (a) the luminance information just one line before the luminance input.

The (f) horizontal edge detection output of FIG.
The horizontal luminance edge / average extraction circuit 11 outputs an edge, which is "1" at the edge of the luminance input and "0" at the non-edge by the operation shown in FIG. The (g) horizontal average luminance detection output of FIG. 7 is an average luminance information output of the horizontal luminance edge / average extraction circuit 11, and by the operation shown in FIG. 4, the (a) luminance input of FIG. It becomes "1" in the part and "0" in the part with low average luminance. The (h) flesh color detection output in FIG. 7 is an output of the flesh color detection circuit 15, and becomes "1" in the flesh color portion of the video input and "0" in the non-skin color portion by the operation shown in FIG. The (i) vertical correlation detection output of FIG. 7 is the output of the vertical correlation extraction circuit 14 of FIG. 1, and by the operation shown in FIG. 5, the (a) luminance input of FIG. "1", and "0" in a portion having no vertical correlation. (F), (g),
The setting values of the threshold setting circuit 12 and the variable attenuator 10 are determined by the signals (h) and (i).

The (j) noise control algorithm of FIG.
The content based on the settings determined in this way is shown. As shown here, noise is not removed in a portion having low vertical correlation and a luminance signal edge portion, and in a portion having low luminance, intra-frame coding (I) forward prediction coding (P) frame bidirectional prediction is performed. Depending on the coded (B) frame, thresholds and gains are switched and applied to remove noise. Further, noise is removed by a gain of 0.3 in a skin color portion having a high luminance, and is removed by a gain of 0.5 in a non-skin color portion having a high luminance. The (k) variable limiter output in FIG. 7 is an output extracted in a form conforming to the above-described noise elimination algorithm. The output is given a gain conforming to the above noise elimination algorithm by the variable attenuator 10 in FIG. (L) Noise output. The noise output is subtracted from the luminance signal by the first subtractor 7 in FIG. 1 to obtain the luminance output (m) in FIG.

That is, in the image noise elimination device according to the first embodiment of the present invention, when mainly removing the background noise, the background noise is eliminated, and at the same time, the outline portion,
Since the signal of the vertical correlation portion and the signal of the high brightness skin color portion are removed from the noise, the outline is not blurred and the minute amplitude of the subject is not lost.

FIG. 8 is a diagram showing a time change characteristic of noise of the video noise elimination device according to the first embodiment of the present invention.

In the video noise elimination device according to the first embodiment of the present invention, a video signal encoded by the MPEG encoding method is recorded on the disk 1. In the (q) coding method shown in FIG. 8, each video frame is coded by any of the coding methods of intra-frame coding (I), forward prediction coding (P), and bidirectional prediction coding (B). Indicates whether it is information. Here, in intra-frame encoding (I), compression encoding is performed using only intra-frame information of the original image,
In forward prediction coding (P), coding is performed using difference information from information predicted by motion prediction from a past frame, and in bidirectional prediction coding (B), coding is performed using difference information from past and future frame information. Is done. Therefore, as shown in the (r) code amount of FIG. 8, the information amount of the compressed frame is the largest in the intra-frame coded (I) frame and the smallest in the bidirectional predictive coded (B) frame. Conversely, as shown in the visual noise amount of the reproduced image in FIG. 8 (s), in the intra-coded (I) frame, while there is much information on components visually detected as noise, In the bidirectional predictive coding (B) frame, there is little information on components visually detected as noise.

As shown in Tables 1 and 2, the threshold is high and the gain is high in the intra-coded (I) frame, and the threshold is medium and the gain is medium in the forward prediction coding (P). Since the threshold value is low and the gain is low in the bidirectional prediction coding (B), the noise removal amount is high in the intra-frame coding (I) as shown in the noise removal amount in FIG. In (P), the value is medium, and in bidirectional predictive coding (B), the value is low. As a result, the amount of noise after noise removal is suppressed from changing with time as shown in (u) Visual noise amount after noise removal in FIG. Therefore, it is possible to eliminate the phenomenon that noise, which has conventionally been a problem, appears to increase or decrease periodically.

That is, according to the first embodiment of the present invention, the horizontal luminance edge / average extraction circuit, the vertical correlation extraction circuit,
By including a skin color detection circuit, an encoding type extraction circuit, a threshold setting circuit, a variable limiter, and a variable attenuator,
By changing the noise extraction sensitivity and the amount of noise removal by various information of the video signal, without eliminating the deterioration of the contour and the minute amplitude of the subject, the noise does not appear to increase or decrease periodically, Provided is a video noise elimination device capable of mainly removing noise in the background.

In this embodiment, the edge of the horizontal luminance,
Although the threshold and the variable attenuator gain were controlled by the average luminance level, vertical correlation, hue, and coding method, this combination is not necessarily limited to the combination and control value described in the present embodiment. Even in a skin color region having a low luminance level, even if the threshold value is fixed at 1, the effect of the present invention is not changed.

Further, in FIG. 1 and subsequent figures, the components following the reference numeral 4 are in the form of a circuit in the present embodiment, but these can be replaced by software.

Further, in the present embodiment, the description has been made with respect to a video signal recorded on a disk medium. However, this is based on an information signal including a video signal of another tape medium or a satellite broadcast or a terrestrial broadcast. Is also applicable.

[0063]

As described above, according to the present invention, noise extraction means for detecting noise of a luminance signal according to a threshold value determined according to the characteristics of an input video signal, and Attenuating means for attenuating the output of the noise extracting means in accordance with the determined attenuation amount, and noise removing means for subtracting the output of the attenuating means from the luminance signal and removing noise contained in the luminance signal. By changing the noise extraction sensitivity and the amount of noise removal, it does not eliminate the deterioration of the contour and the minute amplitude of the subject, and the noise does not appear to increase or decrease periodically. Can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a noise removal device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing characteristics of a first bandpass filter of the video noise elimination device according to the first embodiment of the present invention;

FIG. 3 is a diagram illustrating characteristics of a variable limiter of the video noise elimination device according to the first embodiment of the present invention;

FIG. 4 is a block diagram showing a configuration of a horizontal luminance edge / average extraction circuit of the video noise elimination device according to the first embodiment of the present invention;

FIG. 5 is a block diagram showing a configuration of a vertical correlation extraction circuit of the video noise elimination device according to the first embodiment of the present invention;

FIG. 6 is a diagram showing an operation of the skin color detection circuit of the video noise elimination device according to the first embodiment of the present invention.

FIG. 7 is a diagram illustrating noise removal characteristics of the video noise removal device according to the first embodiment of the present invention.

FIG. 8 is a diagram showing a time change characteristic of noise of the video noise elimination device according to the first embodiment of the present invention;

FIG. 9 is a block diagram showing the configuration of a conventional video noise elimination device.

FIG. 10 is a diagram showing characteristics of a first bandpass filter of a conventional video noise elimination device.

FIG. 11 is a diagram showing characteristics of a first limiter of a conventional video noise elimination device.

FIG. 12 is a diagram illustrating noise removal characteristics of a conventional video noise removal device.

FIG. 13 is a diagram showing a time change characteristic of noise of a conventional video noise elimination device.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 disc 2 pickup 3 disc rotating device 4 video signal reproducing circuit 5 first bandpass filter 6 first limiter 7 first attenuator 8 encoder 9 video signal output terminal 10 variable attenuator 11 horizontal luminance edge / average extraction circuit REFERENCE SIGNS LIST 12 threshold setting circuit 13 encoding type extraction circuit 14 vertical correlation extraction circuit 15 skin color detection circuit 16 variable limiter 31 luminance input terminal 32 first delay circuit 33 second delay circuit 34 third delay circuit 35 first integration Device 36 second integrator 37 second subtractor 38 first threshold 39 first comparator 40 edge information output terminal 41 adder 42 second threshold 43 second comparator 44 average luminance information output terminal 51 Luminance input terminal 52 Line memory 53 Fourth subtracter 54 Second bandpass filter 55 Third threshold 56 Third較器 57 vertical correlation output terminal

 ──────────────────────────────────────────────────続 き Continued on front page F-term (reference) 5C021 PA12 PA35 PA42 PA53 PA66 RA02 RA11 RB03 RB09 SA25 YA01 5C053 FA21 FA24 GA11 GB29 HA06 HA09 KA02 KA14 KA22 5C059 KK01 MA00 MA04 MA05 MA14 PP04 PP16 PP22 PP29 SS02 SS27 TA10 TB10 TD08 TD12 UA13 UA21 5C066 AA05 CA07 EC12 GA02 GA05 HA01 KC04 KD04 KD06 KE02 KE07

Claims (7)

[Claims] 1. A video noise elimination device for removing a noise component of a luminance signal from an input video signal composed of a luminance signal and a color difference signal and outputting the same, wherein a threshold value determined according to characteristics of the input video signal is Noise extracting means for detecting noise of the luminance signal in response to the input signal, attenuating means for attenuating the output of the noise extracting means in accordance with an amount of attenuation determined in accordance with the characteristics of the input video signal; Means for subtracting the output from the image signal and removing noise contained in the luminance signal. 2. The video noise eliminator according to claim 1, wherein the threshold value of the noise extraction unit and / or the attenuation amount of the attenuation unit are switched in accordance with the amount of change in the luminance of the video signal. . 3. The video noise eliminator according to claim 1, wherein the threshold value of the noise extraction unit and / or the attenuation amount of the attenuation unit are switched according to the horizontal average luminance of the video signal. 4. The video noise eliminator according to claim 1, wherein the threshold value of the noise extraction means and / or the attenuation amount of the attenuation means are switched according to the vertical correlation of the video signal. 5. The video noise according to claim 1, wherein at least one of the threshold value of the noise extraction unit and the attenuation amount of the attenuation unit is switched according to the hue obtained from the color difference signal of the video signal. Removal device. 6. An input video signal comprising a luminance signal and a chrominance signal obtained by decoding information obtained by encoding a video signal by three encoding methods, that is, intra-frame encoding, forward prediction encoding, and bidirectional prediction encoding. A video noise elimination apparatus for removing and outputting a noise component of a luminance signal from a video signal, wherein the frame of the video signal is the intra-frame encoding, the forward prediction encoding, and the bidirectional prediction. 2. The video noise eliminator according to claim 1, wherein the threshold value of the noise extracting unit and / or the attenuation amount of the attenuating unit are switched between when the encoding is performed. 7. The noise extracting means includes a band-pass filter means for passing only a specific frequency component of the luminance signal, and a limiter means for comparing an output of the band-pass filter means with a threshold value and passing the signal only below the threshold value. 7. The image noise elimination apparatus according to claim 1, wherein: